Ok, thanks. I can’t really pick a heat sink very quick, too hard to tell if they’ll fit. Thought they’d have some specially designated for that transistor package but, I can’t even tell for sure what package it even is; none of the options for heat sinks even have thevsame numbers as the transistor data sheet.
The transistors I’ll probably just install in-line, surrounded by air. I’ll worry about heat sinks if they get too hot like that, so, I ordered everything else. Thanks again.
So, yeah, I made a heat sink from angle aluminum, bolted the transistors to it with heat-sink paste and mounted it with nylon screws and spacers.
The system has 2 control boxes, upper and lower. The two control boxes have electrically identical circuitry with DPST momentary toggle switches in the bottom box and DPST momentary push-buttons in the top box.
One pole of the switches powers up a coil for a particular hydraulic valve (6 in all: ram 1 extend, ram 1 retract, ram 2 extend, ram 2 retract, motor rotate clockwise, motor rotates counter-clockwise), to select the function, and the other pole powers up the voltage follower that runs the proportional valve for speed control. There is an on-off-on toggle to power up only one box at a time.
The emitters of the tranistors are connected to their respective coils, which are grounded on the other end.
The collectors are all connected to +12V, unswitched.
Each transistor base goes to a node that splits off to the two control boxes. In the inactive control box, the 6 function conductors just hook to an open switch so there should be no interference.
Because of the passive resistors on each side of the potentiometers, I shouldn’t have connected the bases that way but that’s not what’s causing the problem and I know 2 ways to fix it: either move the resistors from after the pots to after the node, or else put a switch instead of the node, to select which speed control goes to the driver base.
Here’s the problem: with the lower control box connected, the system works beautifully. But with the upper control box connected, the lower control box no longer works right, even though the wires only connect to open switches.
Example: if Rotate Clockwise is operated, then some other function, it will rotate clockwise no matter what other function is selected, over and over again. Unplugging the upper control box wiring brings everything back to normal, the lower control box runs the hydraulics perfectly.
It’s like the coils still have current just because of a dead wire hanging off the base node.
I ohmed out the (brand-new) cable and there are no shorts it. Disconnecting one conductor at a time led to this: with the entire upper control box out of the picture, and the other ends just dangling in the air, it still happens. Various combinations yield different crosstalk patterns but there’s no doubt the long wires themselves are making the trouble.
(If I connect just the upper box and disconnect the lower, electrically same circuit, it doesn’t work right, seems to select multiple functions at once.)
For a final verification, I 1) disconnected all upper control box wiring and ran all the functions with speed control, then 2) connected only the upper box speed control wire to its base node (other end floating in the air) and observed the proportional valve remains open even after the function switch is off and the speed knob back to zero. Repeat 5 times, same result.
It’s like the coil will not discharge if there’s a long open wire tapped on to the base of its driver. Also, it seems the long wires render the upper control box unfunctional.
Next warmish day I will connect the upper control box to the shorter wires of the lower control box just to verify it’s actually the same but the ohm meter says it’s right.
The cable is #18 stranded 10-conductor for the signals and #14 stranded for power and ground strapped alongside, approximately 50 feet long. It only carries a few mA for the voltage follower plus the base currents for the selected function and speed control.
The modified voltage follower current goes from +12V, through a 2K resitstor, a 2K pot, then a 1K resistor. The center tap on the pot goes right to the base of the coil driver.
So, why do the long wires cause this problem? It’s just DC so, how can a coil stay charged up after the switches all open up, but only with the long conductors in the picture? What components can I buy to drain off or soak up whatever charges or reflections or whatever’s locking up the coils?
Thanks, Happy New Year.
A few things:
A wiring diagram would be quite helpful–I believe you should be able to post images/PDFs now, and the process of making a diagram (if you haven’t already) is inevitably useful for documentation or troubleshooting at a later date.
On your heatsink assembly, are the transistor tabs isolated from each other? You mentioned use of some insulating hardware, but it’s not entirely clear what you mean. It would appear that things aren’t totally askew there if it’s common to both of your configurations and one works while the other doesn’t, but it’s a possibility one can’t ignore.
50 feet of wire looks like an antenna for RF signals within the transistors’ frequency range, not to mention static discharge and a lot of other stuff that might unexpectedly ruin one’s day… I had no clue you were contemplating wire runs on that order of length. Some sort of protection/filter network would definitely be in order, with something along the lines of the below being a decent starting point; applied at the transistor location, the RC combination should attenuate most RF pickup without messing with your pot setup too badly, while the unidirectional TVS diode should guard against transients of either polarity with respect to ground.
Thanks, Rick_1976, and Happy New Year.
No, the tranistor tabs are not isolated from one another but that’s ok because they are internally connected to the collector pins, which are all connected together with a +12V power bus. I could have used the heat sink itself as the power bus, except my heat sink paste is electrically insulating. The entire assembly is electrically isolated, mounted in the box on nylon bolts, nuts, and spacers.
I think what’s happening is, without the 50’ wire, the bottom controls work great. Then, with a dead 50’ wire connected to a node between the switch and the transistor base, the 50’ wire charges up to 12V. When the switch opens up, the long wire has no place else to discharge but through the transistor base, which current is too low for my amp meter to detect, so the transistor remains “on”. The nature of the hydraulic manifold makes it “pay attention” mainly to the easiest valve, hence sticking to the “rotate” function even when I power up another coil driver, which then also sticks “on”, etc. No idea how long they take to discharge, but it’s several seconds at least. Can’t leave it in this state very long for fear of harming the hydraulics.
The 50’ wire runs inside two 20’ steel square tubes with only a few feet not so shielded (at the ends of the tubes) so maybe the RF isn’t such a problem, though it may be good to filter for it anyway. Maybe a set of grounding straps would also be in order as the tubes are mounted on greasy bearings? However, the entire machine just floats electrically; there is no electrical connection to earth.
The next thing I plan to try is putting coil driver transistors in the top control box as well as the bottom so the base conductors are only a few inches long, and connecting the top and bottom emitters to a node right before the coil. This way the long wire can discharge straight through the coil, which is 5 ohms, not giga-ohms or whatever the input impedence of the transistors is.
The drawback is that the full coil current now has to go through the long wire (2.5A max, 1A typical, 30 seconds max on time, #18 stranded wire). I think it’ll be ok, but I imagine the inside of the tubes gets pretty warm in the hot sun. Maybe 135°F or so. Also the voltage follower may need further tweaking.
Don’t have the new schematic done yet but will draw it up this evening. Did you want to see the old one too even though the circuit didn’t work? Do you think I’m on the right track?
Some sort of wiring fault or unanticipated connection would appear to remain as a possibility, but those are sorta hard to diagnose remotely. Try the filter and see what that does.
Draw the system as it exists–no point in causing confusion by doing otherwise. While I think your present approach is probably a decent one, I’d have no issue with running 2-3A through 18AWG.
Yeah, staring at the old drawing (SharedB) I see the speed control wire should have been able to discharge quickly through the 1Ks after the pots, but even with only that one long wire connected to the node down below and nothing up above it still went haywire so, yeah.
I thought with the new design (RFFiltSharedE), it would be more immune from the interference even without the filters since the long wire no longer inputs to the base, but just carries the output instead.
I had some TIP120s on hand, which I had ordered from some competitor before I found DigiKey, but then your beefier devices got here first, so there they’ve sitten. They fit a little nicer in the upper box, and the function coils only draw an amp or so.
I’ll probably want to mount all these devices onto a couple of cards, rather than floating around in the spaghetti-wires (still in prototyping mode here), and maybe all those spade connectors ain’t so good either so I’ll have to hunt up some cards with matching edge connectors or whatever. It would be really cool if the 2SD2083 all spaced out along the heat sink would plug right in but prob’ly not.
Do you see any more potential problems with this design?
SharedB.pdf (958 KB)
RFFiltSharedE.pdf (1.28 MB)
Found a couple bugs so please throw RFFiltSharedE.PDF away. Does SharedESwitchC.PDF look saner?
SharedESwitchC.pdf (1.37 MB)
-Change only one variable at a time; add the filter to what exists currently, and see if that improves things. If it doesn’t, the problem is probably other than what’s been suspected.
-If you go this different direction, I’d add a diode such as a 6A6-T downstream of the emitter, to prevent a signal applied from one end from reverse-biasing the B-E junction at the other.
-I’d suggest using suitably-rated switches for all the non-proportional loads and getting rid of the transistors & related support components. As drawn, it’s an extra 48 things to go wrong that don’t really buy you any added functionality. Adding a recirculating diode across those loads would help reduce switch wear due to coil inductance.
Thanks. Yeah, I thought the drivers would protect the coils too; they don’t need the whole 12V to pull the plunger. I guess a resistor in front of the coil would do the same thing, lots simpler.
Not familiar with how to add a diode downstream of the emitter, can you diagram it for the proportional control?
Which diode would work best for the recirculating diode?
If those coils are rated for an automotive 12V, they should require no further protection–coils of copper tend to be rather robust.
Think of the diode as a check valve pointing in the direction of the arrow, and add one between each emitter and the proportional coil.
Any diode with a reverse voltage rating adequate for an automotive 12V and a peak forward current rating in modest excess of the coil’s current draw should suffice. Assuming they draw a similar 2-3A to your proportional coil, most anything in the 1N400x family should be adequate.
Thank you,. Hydraforce Tech Support says the coils are rated for continuous at 115% of nominal = 13.8. My system is more like 13.5 but they don’t run continuous. Mostly it’s a few seconds to ≤ ½ minute, similar-length pause, 3-5 cycles, then long rests, even like ½-¾ hour. Maybe 50-100 cycles in a very busy day.
Also, the coils prefer a driver but should operate fine with just a switch, says Tech Support.
Worrying all that switching might be rougher on 'em than running continuous, and averse to running anything right-on-the-hairy-edge like that, I’m hoping a resistor in series might help soak up some of the shocks, and not do more harm than no good.
With a resistor value that would drop exactly nominal 12V across the 10Ω coil, it theoretically still draws 1.2A, only a little more than measured way back when before the original electronics came out.
Do you think this is a good idea?
I’d suggest that short-duration operation results in less thermal stress than continuous operation, in the same way that spending 30 seconds outside wearing shorts and a t-shirt when it’s -20° out is less stressful than doing the same for 30 hours.
Go ahead and add an ohm if you’d like. I’d suggest something rated for at least 3 watts. Depending on how beefy your chosen switches are, I’d again suggest recirculation diodes as suggested above–insights as to why can be found here.
Thanks for the info. Looks like the only way to get the right wattage resistor is to put a few higher-value resistors in parallel. After reviewing your insights it seems like downstream diodes like on the proportional coil would also help protect the switches?
I wasn’t thinking of thermal stress so much as the insulation breaking down within the coils from spikes or surges during switching, is why I wanted to protect the coils. Maybe a non-issue?
Don’t over-think the resistors if you choose to use them; those linked in my last (1Ω/3W) should be suitable if I understood your info correctly.
Recirculation diodes do offer some benefit for reduced voltage stress on the coils, though more so for switch life in terms of reduced contact arcing.
Honestly, I’d expect that things would work just fine for quite a time without either provision, but there’s no harm in making life a bit more comfy for something that would elicit colorful language in the event of untimely failure.
It’s 10Ω-resistance coil, but first they said 13.8V max, then they said 1.2A max. Just gonna tone them down to ≤1.18.
Success! The skybasket is back. I have just flown up pulling vines and scraping their little Cling-Ons from a tall house in the freezing cold. The controls worked smoothly and consistently; no more “soft spots” or “dead zones” along the pot.
Had to re-arrange the resistors, ended up with 4 columns x 2 rows of 10Ω ½Watt resistors to make a 5Ω 4Watt resistance in series with the 10Ω coil to make a nominal 0.9A current. Works lovelyly.
Hafta wait 'til Summer to see for sure if my homemade heatsinks work, but no problems in the freezing cold. I got plenty of spare transistors now since the function coils use current-limiting resistors instead of drivers like the first try.
I got a lot to learn about ordering parts, though. Never imagined the PC boards came in different thicknesses, so the little one is kinda flimsy.
Also, the legs on the solder-on quick-connect blades were spaced properly but too thick for the holes, as were the diode leads, so stuffing the boards was mainly an exercise in fitting fat pegs into smaller holes. Hours and hours of filing, trial-and-error.
They look like they were soldered up by a half-blind kindergartner, but they’re solid, functional, and well-hidden, so who’s gonna know?
The push-buttons (CW Industries P/N GPB527X2SERIES DPST)seem to have 2 positions. Push lightly, they’re momentary-on; push heavily, they’re click-on - click-off. This is very nice for one-handed operation, strictly forbidden by the original design, which forced you to push buttons with one hand and pull the (momentary potentiometer) trigger with the other. Moved the trigger close to the buttons and removed the spring, so I can just leave the pot in whatever position and push whatever button. Love it, love it.
The only disappointment is, still can’t run 2 functions simultaneously, probably because of the hydraulic circuitry. I’m certain 2 function coils power up at once, but only one valve at a time actually responds, oh well. Still a vast improvement.
Never could have done it without you. In other words, DigiKey was the key to this success. I can’t ever repay you for your help and guidance except to spread good word-of-mouth about DigiKey.
I don’t really rub elbows with electronics types that often, but if you send a “Powered by DigiKey” sign I’ll stick it on and fly it everywhere the skybasket goes.
Thanks again for all your help.
